Monday, June 26, 2017

Let's Talk About Self-Driving Cars

The Revolution Next Door
A Primer on Our Driverless Future by Paul Watson

                Cars that drive themselves. It sounds like science fiction, but today, right now, it is science fact. You and I are going to see our world transform over the next few years, and the change has already started. We are about to witness a technological shift that could do for transportation what the internet did for information, a leap forward in transport as big as the jump from carriages to Corollas. Self-driving cars are here. It’s time we started talking about them.
                Most of the driving done by self-driving cars so far has been done by Google’s driverless fleet and by Teslas. Tesla’s AutoPilot program is designed to drive the car on highways, and, while it can handle the car by itself, Tesla stresses that the human in the car has to be ready to take over at any time. By November 2016, Teslas in AutoPilot mode had driven themselves 300 million miles, more than twice the distance between the Sun and Mars. Google has been in the game since 2009, and its driverless cars have driven over two million miles. These cars drive in the city like any normal car, not just on the highway.
                It’s easy to write off robot cars as novelties for rich people or prototypes that will mean something to the next generation. But Ford doesn’t think so. In February 2017, Ford announced that it was investing a billion dollars in self-driving cars. Ford intends to offer a fully autonomous car by 2021. And by “fully autonomous”, they mean no steering wheel, no pedals. General Motors, Nissan, and Uber are investing heavily too, and they anticipate autonomous cars being on the road by 2020. Teslas are too expensive for the masses now, but the price tag is coming down while the technology only gets better. Tesla plans to have a fully autonomous car by 2019.
                Autonomous vehicles have immense potential to change our lives for the better. The most obvious way is by simply not killing us. In the United States alone, 30,000 people die in car accidents every year (worldwide, that number is a shocking 1.3 million), and human error is responsible for about 95% of all crashes. Google’s cars have been in fourteen crashes, and thirteen of those were the other car’s fault. So far, only one person has died while riding in a Tesla, and there is a growing consensus that self-driving cars today are already safer than the average human driver. AutoPilot technology already predicts accidents that humans can’t. And, whereas humans can only learn from their individual experiences, computers can learn from collective experience, meaning that, once one computer driving a car deals with a bizarre occurrence, all the other computers will be prepared for it. Not only are autonomous cars safer than people, they’ll only get safer from here.



                Self-driving cars will save us a lot of time. People are not very good at following traffic rules. They go slow in the fast lane, change lanes without warning, neglect using turn signals, take two spaces to park, and don’t notice when lights turn green. All this wastes time and makes traffic less predictable. Self-driving cars won’t just make our commutes shorter and easier, but they will set our commute time free. Instead of focusing on the road, we can use commute time to prep for a meeting, watch the news, or catch up on sleep. Our cars will become less utilitarian machines and more extensions of our homes and offices. They will drop us off and pick us up, and they could drive themselves home in between, eliminating that stressful search for a parking spot. Autonomous cars will be like a personal valet available anywhere twenty-four hours a day. People groups that depend on others to drive them around, like children, seniors, and the disabled, will be granted their travel independence. Drunk drivers will be a thing of the past, because no one will be driving himself home, drunk or sober.
Long drives that eat into our business trips and vacations can instead be started the night before. A passenger could get in his car at 10PM in New York, enter the destination, sleep during the trip, and arrive at 8AM in Chicago the next morning, saving on time, air fare, and hotels. Even overland freight will be better, faster, and cheaper. Currently, there are strict regulations on how much time truck drivers can spend on the road in any given day or week, but a robot can drive across the country without breaking to eat or sleep.
All of this will save a huge amount of money. Auto collisions cost half a trillion dollars worldwide every year. Cars that don’t crash don’t need to be replaced. Fewer accidents means much less liability insurance. Cheaper and faster freight will make business operate more efficiently and save consumers real money at the checkout.
Of course, this won’t all happen immediately. These things will be true at some point in the near future, but the change will be gradual. And there are risks. Computers are susceptible to viruses, hacking, and simple system failure. We’ve all had a system update that didn’t go quite right or a maddening malfunction that seemed to be caused by nothing. The idea of something like that happening to your car, a three-thousand-pound hunk of metal that can travel at a quarter of the speed of sound, is terrifying. Also, the short-term future of self-driving cars will certainly involve car-to-car communication through some sort of central system. What happens if this system is attacked by cyberterrorists? Will one bad actor with a computer be able to crash every car on the streets with a few lines of malicious code? There are ways to address these problems. After all, our current computer systems tend to operate very well and have adequate security. But these are real problems that have to be overcome before our society can make a full transition.
There is also a widely-held concern that the automation of transport will lead to massive job losses. Truckers, cabbies, and delivery drivers will all be put out of work by self-driving cars. Domino’s already has a robot that delivers pizzas, and Amazon is famously working on drone deliveries so they can cut out UPS. Auto insurance providers will see demand for their business erode. Airlines and hotels will lose business-class customers, a major part of their clientele.
Sound economics says that these kinds of job losses are a natural part of the economy becoming more efficient. After all, there used to be factories making whips for carriage drivers to use on their horses. Those factories and the jobs they created disappeared because they stopped being useful to society, and the people who worked there got productive jobs that helped the economy and made the country richer. But many people worry that so many workers suddenly losing their jobs and being forced to start new careers could put a massive strain on the economy.
The change is already underway, but it will be more an evolution than a coup d’car. At first, cars will have normal human controls with an additional autopilot feature. Humans will be able to take over optionally or in emergencies, a lot like how cars are depicted in the movie I,Robot. This is essentially the stage we’re at now. As mentioned, Teslas drive themselves on highways all the time, and human drivers take over when necessary. More and more of these cars will be on the road as time passes and as the technology becomes better and cheaper.
Before long, having a self-driving car will be like having a color TV in the 60s. Adoption of the technology will happen at a faster and faster rate, until it’s more like having a color TV in the 90s. We could see driverless cars become the majority as soon as 2025. There will still be human-driven vehicles in rural areas, farms, and construction sites, but, for most people living in urban areas, we will reach saturation, where practically all cars on the road will be driven by robots instead of humans.
Over time, it is likely that the concept of individual ownership of cars will fade into history. As cars become less dependent on people and more dependent on each other, cars will become units of a single cooperative system rather than independent agents. Currently, most cars are inactive most of the time. They would be more useful if they were shared. Some people suggest that cars in the future will link with other cars to form something closer to a train, except that the individual units will be able to separate far more easily. When the technology reaches this stage, autonomous vehicles may be so efficient and versatile that they replace all public transportation.
Considering the benefit that autonomous cars are capable of, especially the immediate promise of saving so many lives, some people think it’s unethical to delay. It seems obvious that it’s important to get the technology right and safe before it diffuses across society, but it’s arguably already there, and every day more people die simply because humans are bad drivers. Have we already passed the point where excessive caution has doomed people that would otherwise be alive today?
This is the first of a whole body of ethical questions surrounding self-driving cars. Another question deals with how cars should be programmed. Should a car be designed to kill its own driver rather than hit another car with two passengers in it? It seems obvious that cars should be programmed to save the highest number of people possible, even if that means occasionally taking one for the team. But would people be willing to drive cars that were programmed to value all lives equally rather than defend their own? And should this be a top-down decision or a bottom-up decision? After all, if all cars are designed to save the greatest number of lives, but you decide to program yours to save your life no matter what, you have increased your chance of survival at the cost of increasing everyone else’s chance of death. This is a prisoner’s dilemma, a situation in which all parties benefit from cooperation, but each individual stands to benefit more if he, and only he, cheats. Typically, in such situations, all parties cheat, and everyone is worse off. So should this be a decision that the government makes for us? Or auto makers? Or could this be done organically by auto insurers offering discounts to drivers who elect to have their cars programmed to save the greatest number of lives? These are some pretty sticky questions, but, whatever is decided, society will still be far better off due to the reduction of harm caused by having humans behind the wheel.
Although they will be safer than human-driven cars, self-driving cars will still occasionally cause accidents that cost lives and destroy property. When a human driver causes such a crash, they are held legally responsible. But who is at fault when a computer is making the decisions? It’s easy to draw an analogy between this and holding people accountable for their animals. If your neighbor’s dog jumps the fence and bites off your thumb, you are entitled to compensation. On the surface, it seems responsibility for self-driving cars should work the same way. But it’s also conceivable that a driver could claim faulty programming by the manufacturer is to blame. Volvo and Mercedes have indicated they may accept that responsibility, but what if they are the only ones? There is potential for a seriously messy legal issue here.
While it’s inevitable that self-driving cars will saturate the roads in the near future, there will still be people who refuse to stop driving. When, if ever, should driving your own vehicle be banned? Could using a steering wheel be a crime in the future? The more self-driving cars there are on the road, the safer everyone is. In a sense, a driver who chooses to drive himself is choosing to make people around him less safe. Maybe this problem will solve itself. Perhaps so many people will willingly switch to autonomous cars that it’s no longer economical for auto makers to make cars with steering wheels. Maybe the holdouts will be forced to change simply because human-driven cars go obsolete, like flip phones or VCRs. But if they don’t go obsolete, is it acceptable to make them illegal for the benefit of society?
It’s unclear how these issues will or should be resolved. But the cars are already here. We live in a cultural moment where every facet of society is politicized, from the clothes we wear to the stores we shop at to the music we listen to. But this hasn’t happened to self-driving cars yet. This is one of the few issues that hasn’t been forced into a false binary between red and blue. Politics is strong enough to destroy or delay the shift forward to a future of self-driving cars, or at least to set a large part of the population against it as a manifestation of their tribal identity. We can’t let this happen. This is a technology that promises to make all of our lives better. It’s a step towards a better future for everyone, rich or poor, in every corner of the globe. The conversations that you and I have today will shape the way our world looks tomorrow as self-driving cars become an increasingly central part of our lives.  Autonomous vehicles have the power to alter the future. So do you. 

No comments:

Post a Comment